Spatial Statistics
  

Ecologists are very interested in the spatial patterns that they observe in nature, because the existence of particular patterns can be used to infer important underlying processes.  For example, seedlings or young trees in a forest are frequently found have a clumped distribution, which may be attributed to a clumped distribution of seed, but large trees are frequently evenly spaced and this is attributed to competition been trees as they grow.  However, the ability to make these conclusion depends upon using statistics to analyze the observed pattern. The spatial dependency within the data makes statistical inference more difficult. Our recent work has focused on evaluating both the performance and ecological examples of two types of spatial statistics: point pattern and lattice.

Point Pattern Analysis

In recent work (Loosmore & Ford 2006) we have shown that point pattern spatial statistics frequently used by ecologists, such as the G (nearest neighbor) and K (all plant-plant distances) are wrongly used when making statistical inference. This is particularly important when attempts are made to assess the scales at which patterns are thought to occur.  More appropriate methods are described in the manuscript and sample software for R is available here.

Additionally we have done significant analysis of the WRCCRF stem map (see Fig 1, above) in an attempt to evaluate the dynamics of forest succession. The data and sample scripts for analysis of this data are available here.

Lattice Analysis

In current work Derek McClure is analyzing the spatial distribution of catches of the northern flying squirrel made in regular arrays of traps and examining how catches can be related to patterns of vegetation.  This work requires determining the power of estimation of Moran's I statistic and developing and comparing different models for spatial dependence between animal captures and vegetation.  This work is important because the northern flying squirrel is an important prey species of the endangered northern spotted owl and managers would like to know what vegetation the squirrel is preferentially associated with.